Jev dictation mode check

Distinguish dictated text from an editor command

Give the transcript and speaker's stated intent. Compare words meant for the document with an instruction to operate the editor, leaving ambiguous command-like text for clarification.

Keep dictated content separate from editor commands.Live API
Try an example

Original synthetic input. Edit the task or evidence to make a new API request.

Source: Workflow inspiration on X

Source and example scope

The inputs, rubric and preview are original. This window demonstrates the stated decision only; it does not reproduce the source application or its measurements.

Choose handling for this supplied transcript using the stated speaker intent.

Decision rules and labels
Insert textEditor commandClarify intentMissing transcript
A real response from the Jev API

Edit any input to explore a different decision. No request is sent until you run it.

Your next decision

Not run yet

Choose an example or write your own input, then run it to see the ranked choices.

Results are not stored by this site. Your submitted text is processed by the API provider.

The decision

How does Jev dictation mode check use its evidence?

Jev dictation mode check uses a supplied transcript and speaker intent to decide whether text should be inserted or treated as an editor command.

The source describes a speech-to-text application with Jev decisions. Our original demonstration starts after transcription. It does not capture audio, evaluate recognition quality or reproduce that application's operating-system support. The TypeSafe API handles only the bounded text decision.

Words such as select all can be either dictation or a requested operation. The provided intent distinguishes those uses. The local consumer prints a mode report and performs neither the insertion nor the command, so command-shaped content can be tested without changing a real document.

Original Jev dictation mode check workflow diagram showing supplied evidence, a typed decision and the local preview.
Original workflow illustration. The output is a local preview, not an executed action or source screenshot.
How it works

Try the example in four steps

Use the sample to understand the rule, then test your own input for dictation intent.

  1. Illustration of adding the text or evidence for a decision

    Add the information to check

    Give the transcript and speaker's stated intent.

  2. Illustration of defining a question and its allowed choices

    Confirm the question and labels

    Open the decision rules and keep the intended distinction between Insert text, Editor command, Clarify intent, Missing transcript. Edit the choices only when testing your own rule.

  3. Illustration of sending one input and comparing the returned choices

    Run the decision

    Use a sample first, then try your own input. Compare the selected label with the other returned probabilities.

  4. Illustration of reviewing a result and an uncertain item before using it

    Review the next step

    Review Missing transcript before using a result. The label is a decision about the supplied information, not an executed action.

Our API test

What the six original inputs returned.

What the six original inputs returned.
Original test inputObserved labelLeading probability
Ordinary document dictationInsert text78%
An explicit editor operationEditor command50%
Command words used as document contentInsert text60%
A command-shaped fragmentClarify intent71%
No speech textMissing transcript99%
An explicit undo operationEditor command85%
  • Measured adapter round trips: 481-1244 ms. Estimated input cost: $0.00002159-$0.00002205 per request. These observations are not a latency guarantee or invoice.

6 actual Jev API calls on October 1, 2026. Original synthetic inputs check this rubric and handoff; they do not establish held-out accuracy. Probabilities are model outputs.Verified

Understand the results

The meeting fixture gives ordinary prose and explicitly asks to insert it

The meeting fixture gives ordinary prose and explicitly asks to insert it.

Read the example notes

Insert text describes that mode. The demo keeps the complete supplied sentence but does not generate document text or edit a file. Recognition errors in the upstream transcript would need their own evaluation.

Source and example scope

The select-all fixture explicitly requests an editor operation. Editor command records that intention without performing the operation. A real executor would need an allowed command mapping, focus checks and authorization; the model label alone should not determine which window receives keyboard input.

The quoted words belong in the document

The quoted-content fixture supplies the same words as the command fixture but states that they belong in the document.

Read the example notes

Insert text tests whether the stated intent survives a familiar command phrase. This is an original functional comparison, not a production study of all speech-command ambiguities.

The delete-that fixture provides no distinction between content and an operation

The delete-that fixture provides no distinction between content and an operation.

Read the example notes

Clarify intent preserves the gap before an application acts. It does not assume that a short imperative is always a command. A speaker's clarification would need to travel with a new request.

The absent fixture supplies an intention to dictate but no transcript

The absent fixture supplies an intention to dictate but no transcript. Missing transcript differs from a complete phrase with ambiguous handling. The local preview has no text to insert, and the demonstration cannot recover audio or invent what the speaker said.

The undo fixture requests a different editor action explicitly

The undo fixture requests a different editor action explicitly.

Read the example notes

Editor command checks the general mode boundary without mapping a specific shortcut. A production speech interface should separate mode selection, deterministic command parsing, user confirmation where needed and actual document modification. Each step needs its own current evidence.

Using the result in your application

Download the exact sample requests and saved observations below.

Read the example notes

Expected labels are original fixture checks and stay outside the request sent to Jev. The observation file preserves the complete probabilities and the hash of each exact request. Editing a task, rubric or candidate requires a fresh response; a matching sample name does not make an older answer current.

Using the result in your application

The JavaScript runner makes one independent SDK request for a chosen sample ID.

Read the example notes

It reads TYPESAFE_API_KEY privately from the environment and has no automatic retry. The Python consumer operates offline: it checks the rubric, labels, unique sample identities, request hashes and valid probability distributions before selecting a fixed local preview. A tied leading result remains pending for review.

Source and example scope

The page reports measured adapter round trips and estimated input cost for the recorded calls.

Read the example notes

Those values describe these requests, not a provider latency guarantee or a billing statement. Six deliberately authored fixtures can check the stated behavior, but they do not establish accuracy on an unseen dataset. Keep disagreements and missing-input examples when testing an integration.

Runnable offline example

Replay one request and inspect the saved preview.

View the code example
Replay commands
npm install @typesafe-ai/sdk@0.6.0
# Set TYPESAFE_API_KEY privately in your shell.
node run-choice-sample.mjs samples.json content
python preview_choice_workflow_v2.py samples.json saved-api-results.json
Scope of this demo

Where does this decision fit?

Useful with explicit supplied evidence

  • Distinguish a supplied speech transcript intended as document content from an explicitly intended editor command, keeping ambiguous intent pending. Text-only original transcripts; no audio recognition or keyboard operation.
  • Inspecting the original task and a fixed output before integrating external tools.
  • Comparing an explicit expected label with the complete recorded distribution.

Needs separate implementation or evidence

  • Inferring facts absent from the supplied input.
  • Treating a semantic selection as authorization to act.
  • Executing an external action or generalizing six fixtures into an accuracy claim.
Example FAQ

Questions about this original decision workflow.

Keep relevance, compliance and execution separate.

Where does dictation mode checking start?

It starts with a supplied text transcript after speech recognition. This page does not record audio.

Does Editor command run a shortcut?

No. The local report identifies a mode. Allowed command parsing and execution are separate.

Can command words be inserted as prose?

Yes, when the supplied speaker intent says they are document content.

Why keep Clarify intent separate?

Command-like wording alone can be ambiguous. The rubric preserves an unspecified content-versus-operation intent.

Do these tests measure speech recognition accuracy?

No. The transcripts are original text fixtures and no audio model is evaluated.

Try a task from your own workflow

Change the supplied task or evidence here, then inspect the live response. Playground supports another explicit decision.